Paul Lauterbur and the Invention of MRI

by M. Joan Dawson

  Paul told the Dow Corning chemists something almost unbelievable: he would obtain the NMR signals from silicon, the life blood of the company! He proposed to look at silicon compounds from the inside, straddling the critical silicon atoms. He had calculated that NMR signals could be obtained from 29Si, and all could see that this would be enormously important. And that was why the Dow Corning group wanted an NMR spectrometer, and that was why Earl Warrick and Dow Corning needed Paul Lauterbur.

  What a pair Lauterbur and Warrick must have been! Warrick was brilliant and practical, while Lauterbur was not interested in products or money: he was searching for the soul of nature. Paul was now a pioneer’s pioneer; NMR was still a very new technique in chemistry, and it was almost all accomplished using the hydrogen atom, whose large signals yielded useful chemical data. The young Lauterbur was on fire, his enthusiasm for breaking new ground energizing his colleagues. Earl, intimately familiar with Paul’s ways, was willing to take a chance on the visions Paul put before him.

  Figure 4.1

  Earl Warrick. Courtesy of the Dow Corning Corp.

  Things moved apace. Varian was contacted, and a contract was signed for an NMR instrument.4 When he went to Varian’s test laboratory to evaluate the new spectrometer, Paul took a few samples of common silicon fluids. He knew that finding the signals from his samples would be difficult because they were only about a third of a percent of the proton signal from water. It takes little imagination to realize that great experimental technique was required. When I asked Paul some fifty years later how he was able to do this thing that nobody thought possible, his response was, “Something was always going wrong. I just don’t remember conditions when they went right.”

  A big trouble was that the magnets were very sensitive and became unstable with fluctuations in temperature. People were running around opening and closing windows to “control” the ambient temperature. Varian engineers were using the magnetic signal from 17O as a test of the quality of their machines. Paul once found them in a tizzy because they couldn’t find the 17O signal from water. At that time the various plugs and sockets on the NMR machines were all compatible, and the experimenters had reversed the input and output cables,5 so signals could not possibly be found. Varian and other manufacturers later made all their sockets and cables for input and output incompatible so that this kind error couldn’t happen again. The customers were grateful, but the engineers were just terribly fed up with such stupid, if funny, problems.

  Paul finally observed 29Si signals with the expected characteristics,6 and the machine was accepted. While he waited for the new spectrometer Paul collaborated with members of the Pittsburgh Plate Glass group at Mellon Institute to publish a preliminary 29Si paper.7 This was Paul’s second publication (the army work didn’t come out until later), and it attracted notice. In these studies, and in his later search for even more difficult signals, Paul had to optimize everything about the experiments. For this he used some unconventional techniques,8 and the data he obtained looked unusual, confusing some chemists. So, while everybody was talking about Paul’s work, not everyone accepted it. The NMR signals Paul had searched out were so very small that some chemists, already skeptical about the possibility of finding them, complained that you couldn’t be sure they were real. Paul’s work, as usual, was way ahead of its time. It wasn’t until the 1970s and later, after many technical advances, that 29Si NMR became widely used for studies of silicones and silicates.

  The Jackpot: 13C NMR

  People who do MRI seem to think Paul’s life began the year he invented it. But in the 1950s and 1960s Paul’s spectroscopic work was well recognized. Particularly well respected was his important breakthrough in showing that the ubiquitous carbon atom could be studied by NMR. Most of the carbon in the world is 12C, but about 1% is the stable, NMR-visible isotope 13C, and this was enough to open a whole new world for studies of biology and organic chemistry. Paul’s name was so closely associated with 13C NMR that even into the 1970s, after imaging was born, a student of Paul’s could say, “I’m working with Paul Lauterbur,” and the response would be, “Oh, you’re working on 13C!” This work of Paul’s is almost forgotten now, but there are still chemists around who remember Paul as the “father of 13C NMR,” and they almost universally believe him to be the sole initiator of this field.

  There were actually two “fathers of 13C NMR,” working independently and publishing almost simultaneously, in 1957. C. H. Holm, working at the Shell research facilities, obtained 13C data at almost the same time as Paul. The two studies even appeared in the same volume of the Journal of Chemical Physics.9 Holm did not publish again on 13C, while Paul continued to publish extensively.

  Here’s how 13C NMR happened. While waiting for delivery of his new spectrometer to the Mellon Institute, Paul compared the expected spectral characteristics of 13C with his 29Si data, and calculated that the difficulties in detecting one nucleus were offset by the difficulties for the other. If he could do 29Si, he could do 13C. If silicon was a big deal, then carbon was a really big deal. For all of Paul’s and his employer’s love of silicon, carbon is much more important in our industries and in biology. Carbon is the key structural element in all organic chemistry and all of life. But, as with silicon, scientists generally thought NMR studies of carbon couldn’t be done; in fact, they imagined it so hopeless that they didn’t even think about thinking about whether it could be done. The “I am here” signal would be just an itty-bitty little thing (about 6,000 times smaller than that of hydrogen); just how crazy could Lauterbur get? But carbon and silicon are sisters, adjacent in the same column of the periodic table, and they have similar nuclear characteristics, so Paul went ahead and tried.

  When his new machine arrived at the Mellon Institute, Paul didn’t wait for the installation engineers but impatiently did the initial installation himself (probably voiding all warranties in the process). He had to work at getting conditions right to find signals from carbon, and from today’s perspective, all of his techniques were astonishingly primitive. Paul warmed up the 13C sample in his hand just enough that as it cooled, it offset some of the machine’s temperature-dependent electronic fluctuations. This kept the baseline from scooting off the chart.

  Figure 4.2

  Early shimming methods. From Daniel D. Traficante, “Carbon-13 NMR Spectroscopy: It Wasn’t Always Easy,” Concepts in Magnetic Resonance 3, no. 1 (January 1991): 13–26. Reproduced by permission.

  Among experimental techniques NMR spectroscopists wouldn’t think of today is that Paul put a reference compound, carbon disulfide, into the spectrometer and turned the oscilloscope that displayed the result to maximum. He marked the position of the signal with a red grease pen before it faded away. Then rapidly, before anything could change—magnetic field drift was a significant problem in those days, before field and frequency locks became available—he replaced the reference with the experimental sample, and the position of this signal was also marked with the red grease pen. He used a plastic ruler to measure the distance between the signals.10 It wasn’t quite a black art, but something close. He needed a huge amount of sample, and had the glass shop “supersize” the sample tubes (to 15 mm diameter). He later had tubes especially made with a ground glass joint at the center of the top, so that a standard 5 mm tube containing a reference compound could be centered inside.11 It was a tour de force, not at all practical for an everyday analytical laboratory.

  Paul always remembered the feeling he had when he first viewed the little blip of a 13C signal on the oscilloscope screen. He knew it must be in there, but to actually see it, to prove he could really see it, lifted him, sent him soaring. Later he said, “The emotional reaction was like putting a coin in the slot machine and having the nickels all come falling out.” But Paul’s idiosyncratic technique made the data look funny, not at all what people expected.12 They were not “serious” spectra, some thought. How could you be sure such weak signals were real? This was a valid question. We, at least biologi
sts, have now got used to these weak lines of data because they are often all we can get. We have statistical methods to test the probability that a signal exists and what its characteristics may be. Paul’s contemporaries in the 1950s had no such advantage.

  Paul’s breakthrough produced some heavy weather. When he gave his first talk on 13C NMR, to the National Research Council in Ottawa, John Waugh, then an assistant professor at MIT, made a comment from the audience that Paul’s results were incompatible with some proton studies done at Harvard.13 “When a Professor at MIT says a graduate student is wrong it is cold sweat time” Paul groaned. He returned immediately to the lab to review his notes and repeat his experiments. He realized that he and the Harvard people were actually reporting something slightly different. (The Harvard people must have been measuring from the center 12C to its first satellite, which is one-half of the actual coupling constant, and that was why the numbers did not agree.)

  Despite the grumbling, some excellent scientists could see the inherent value in the stuff Paul was talking about, and this became clearer as time went on. Paul studied over a hundred compounds using 13C NMR, showing smooth curves as he changed from one molecule to its close neighbors. According to Charlie Slichter, “Those smooth curves he obtained showed that there are a few simple parameters that govern molecular structures.” This was an important first step in determining the electronic structure of molecules by NMR, a feat that could not yet be accomplished. Paul said of his first 13C studies, “Pretty much they were in the category of answers to questions nobody asked,” but one of them was so frequently cited that it was designated a Citation Classic by the Institute for Scientific Information. Slichter said of Paul’s first presentation, “I thought the scientific content was just really outstanding, pioneering and original and elegant. He was simultaneously proud and modest. He knew he had something important, but he wasn’t stuck on himself.”14

  This was still NMR prehistory. The equipment was finicky and difficult, even for standard experiments. Things slowly improved, with Paul a leader in the crusade, but for about twenty years, there was only a little follow-up of Paul’s work by other scientists.15 Few scientists had the equipment or expertise to reproduce Paul’s early studies using 13C NMR, and fewer had the imagination or courage. Then, during the late 1960s and early 1970s there came some important breakthroughs, in the form of superconducting magnets and Fourier transform techniques, that made all kinds of NMR spectroscopy much easier, and the field grew quickly from that time. The spectroscopists who did take up 13C NMR in the early 1970s looked in amazement at what Paul had been able to do with his primitive equipment in the 1950s. The data that Paul had obtained were essentially the same as those obtained later with much more powerful technologies.

  After his first big breakthrough, Paul settled cozily into the lab to do a series of 13C NMR studies of aromatic compounds (compounds containing special rings of carbon atoms). These, he believed, as did many others, were the place to start in understanding the NMR characteristics of organic compounds since series of substituted aromatic rings, such as benzene, were easily available. There was a great deal of interest and a general feeling of anticipation with regard to the magnetic interactions (coupling constants) between different species of nuclei in those days. It was hoped that understanding their nature would lead to better understanding of the structure of molecules, so this Paul pursued. He obtained 13C NMR spectra of compounds of substituted carbon moieties and produced smooth simple curves of both chemical shifts and coupling constants that indicated important parameters that determine chemical structure.

  Paul made many more contributions to the new field of science he had created. One was an NMR study of gases and liquefied gases (unpublished) in collaboration with Andrew Patterson Jr.’s group at Yale University, along with an attempt to detect bicarbonate (H2CO3 1-)in solution. Patterson’s graduate student, Ray Ettinger,16 brought materials prepared at Yale to Paul’s laboratory. Paul, who had not yet earned his own PhD, was appointed an outside reader on Ettinger’s resulting PhD dissertation.17

  During the same period Paul was asked to write a chapter reviewing heteronuclear NMR for a two-volume compilation titled Determination of Organic Structures by Physical Methods. The author list bore the names of the most outstanding figures in the field, highly respected scientists. Paul was certainly the only graduate student among this august company. Paul’s chapter, on heteronuclear NMR,18 was the longest thing he ever wrote without a collaborator. He explained clearly the case for NMR, and heteronuclear NMR in particular, emphasizing the directness and simplicity of the spectra. Paul was thirty-three years old at this time.

  Other Fancy Stuff

  Paul continued his high level of scientific productivity across broad areas of interest, concentrating on at least three things at once: the theory of chemical shift, its study by heteronuclear NMR (especially 13C), and chemical shift anisotropy, particularly in solids. The unifying goal was to test theories of electronic structures in liquids and in solids. All of Paul’s work at this time, almost outrageously disparate, was totally connected in his mind. He teased out molecular structure using NMR techniques, and developed new techniques for further teasing. In addition to his work on 29Si and 13C, he obtained the first spectra of aluminum (27Al) and Tin (119S). He also studied cobalt (59Co) and lead (207Pb).

  Paul enjoyed collaborating with John Burke to show the feasibility of NMR studies of tin.19 Burke was a fellow Pitt graduate (in chemistry) who worked as a technician in Paul’s laboratory. According to Paul, Burke was a wild, red-haired Irishman, a self-described hellion in high school. As a result, the only job he could find after high school was stacking boxes in a warehouse. The young Burke quickly decided he wasn’t going to do that all his life, and joined the army, where he got a BS degree, married, and had kids. He returned to Pitt for graduate school and eventually became a major executive in a chemical firm.20

  A concurrent study that greatly satisfied Paul was his work with Bob Kurland, “On the Signs of CH and HH Coupling Constants,”21 published in 1962. Kurland was at that time at Carnegie Tech, and later moved to the University of Buffalo. The size of proton-proton coupling constants became important as a measure of molecular conformation following a theoretical calculation by Martin Karplus. For a long time, spectroscopists did not know the signs of the coupling constants, and it was a big issue. “These went from being unknown to wrong,” Paul said.

  Paul was interested in the phenomenon of chemical shift anisotropy, which is that chemical shifts of a solid can vary depending on the direction in which they are sought, because the electronic structure experienced by a nucleus is different in different directions. A great deal could be learned about the structure of a solid by observing its chemical shifts as a crystal sample is reoriented in the magnetic field, but often the resonances of the individual nuclei are very broad. Paul liked challenges, and wound his mind around this problem. In a pioneering paper in 1958, Paul made the first observation of 13C chemical shift anisotropy in crystals. It was calcite (CaCO3)—an ideal sample, as he pointed out, because all magnetic nuclei are present in such low abundance that dipolar broadening is negligible, and a sharp, strong line is observed.22 On the other hand, Ca-carbonate has a signal decay time of about forty minutes—not very helpful! (The eminent NMR physicist Alex Pines did a similar study on powdered Ca-carbonate many years later with much better equipment, and was much surprised to find Paul’s earlier work.)

  There were other reasons that Paul’s wife, Rose Mary, didn’t see much of him during this time. Paul became a member during the late 1950s of the NMR subcommittee for the American Society for Testing and Materials (ASTM). He was later made chairman. The task of the committee was to find standards for the presentation of NMR data in order to avoid in the future the confusions that were already taking hold, because different laboratories were presenting their data in different ways.23 People were very emotional about which direction and which scale would be used. The formidabl
e (and famously cantankerous) Herb Gutowsky was a member at that time. In response to what Paul believed to be a noncontroversial issue—should the NMR peaks be pointed up?—Gutowsky was the sole committee member who voted to point them down.

  Big Enough for a Conference

  By the late 1950s it was clear that NMR provided key information about molecular structure that could often be interpreted more readily and unambiguously than other analytical data. There were enough practitioners now to feel the need for a specific national meeting devoted to experimental NMR methods. The first of these conferences was a small, informal, one-day meeting in June 1960 at the Standard Oil Company Research Laboratories near Cleveland, run by Bill Ritchey, the resident NMR spectroscopist. There were forty-two attendees, and all felt they had greatly benefited from this opportunity to discuss their common technical problems.

  The group at the Mellon Institute invited those assembled to hold a similar gathering the following year. The Second Conference on Experimental Aspects of NMR Spectroscopy convened at the Mellon Institute on February 24–25, 1961; this time there were 118 attendees. Once again the meeting was about how to deal with technical problems, such as maintaining proper coolant pH so algae wouldn’t grow and clog the machine.24 Paul chaired the fifth such conference in 1964 and replaced a series of long, confusing names with the simple “Experimental NMR Conference” (ENC), an annual event that now draws upward of 1,500 attendees annually. Now people wanted an annual event, and the Mellon Institute hosted conferences from 1962 to1970. They soon had to limit attendance so the crowd wouldn’t be too big for the Mellon facilities.